Thursday, September 1, 2016

The Shift in Violent Crime Rates.

Two years ago I ran across a statistic stating that murders in Washington, DC had declined by 87% from their peak in 1991. Having lived in DC during the eighties in what my landlord referred to as "an aggressively urban neighborhood," I found this intriguing and decided to look into whether it was a general trend across the country. I found that, in the same time period, for the whole of the United States, murder rates dropped 52%. I tried and failed to get someone in the Public Health school interested in assembling a paper on this. Now, two years later, I decided to look in depth into what's going on.

Some of my findings were surprising. In spite of violent crime dropping by 50% across the United States, violent crime has been going up in the border states—the Canadian border states. It has fallen significantly in the four states bordering Mexico. I decided to test a few theories as to what is going on.

A Word About Sources.

The FBI maintains a database called Uniform Crime Reporting which looks at the numbers and population frequencies of violent and property crime. I downloaded the crime statistics by state (and the District of Columbia) going back to 1960 and started digging. [Excel files are available by request.]

First of all, I focused on crime rates (per 100,000), rather than total crimes. California should not be considered the most crime-ridden state because it has six times as many people as the "average" state. On the other extreme, Detroit should not be considered to have a drop in crime rate because of losing population.

Note: the FBI database has some quirks and limitations. The previous year's statistics are not released until late September, so the final year figuring into my calculations is 2014. The 9/11 attacks were excluded from the violent crimes statistics for 2001 which would certainly have made New York City's statistics look bad that year. This exclusion was not extended to other terrorist attacks (e.g., Oklahoma City, Fort Hood). In 2012, the definition of rape was changed which has lead to more crimes being considered rape (38.6% more rapes in 2014 using the new classification versus old). The FBI, however, also reports the number of rapes by the "legacy definition," which I used in my calculations to compare rape figures and to adjust the total violent crime figures of the last two years.

Information on the population make-up of states came from the US Census Bureau. Some were available through their "Quick Facts" section, others as part of the 2010 Census Report.

Violent Crimes, the United States as a Whole.

Violent crimes are defined by the FBI as murder and non-negligent manslaughter, rape, robbery and aggravated assault. Robbery involves the victim being present whereas burglary is when the victim is absent, the latter being a property crime. Aggravated assault and robbery dominate the statistics.

Category           Percent Total
Aggravated Assault    63.6%
Robbery               28.0%
Rape (legacy definition)      7.2%
Murder, non-negligent manslaughter     1.2%

Due to the fact that murder is such a small portion of the overall violent crimes, I decided to mostly focus on the bigger picture, violent crimes.

Violent crime rates rose drastically from 1960 to 1980 (270.8%), rose somewhat from 1980 to 1991 (27.1%), and then began a drastic decline since (down 51.8%).
Violent Crime Rate per 100,000, U.S., 1960 to 2014, from FBI, Uniform Crime Reporting Statistics.
Murder and Non-negligible Manslaughter Rate per 100,000, U.S., 1960 to 2014. Ibid.

Individual years have blips. These can be particularly significant when it comes to sparsely populated states and those with low crime rates. For example, in 1994, North Dakota had one murder, (further evidence that the Coen brothers misnamed their 1996 film) while averaging six a year for the decade. To smooth out these blips, I averaged three consecutive years together with the reported year being the central one.

The year 1992 was chosen as the reference year because this set of three years (1991, 1992, 1993) is when the violent crime levels peaked nationwide with 758.2, 757.7, and 747.1 incidents per 100,000 population, respectively, averaging out to 754.3 for these three years. The three most recent years, 2012, 2013, 2014 saw violent crimes at a rate of 386.9, 369.0 and 365.6, respectively for an average of 373.8, for a drop of 50.4%. (I will refer to these two time points as [1992] and [2013].)

Following this trend, murder rates for all of the United States have declined from 9.5 per 100,000 in [1992] to 4.6 per 100,000 in [2013], a drop of 51.6%.

The States where Violent Crime has Gone Down the Most (and Least).

Below is a map showing the change in violent crime statistics by state from 1991 to 2013. With 50 states plus the District of Columbia, the total of 51 was divisible by three and I highlighted the 17 best performing states (with the greatest reduction in crime) in lime green, the 17 worst performing in red, and the middle group in pink.

Changes in the Rates of Violent Crimes, [1992] to [2013]. The more negative a number, the greater improvement.

These are the ten states with the most improvement.

Most Improved  Percent change
 1 New York     -65.2%
 2 California        -63.3%
 3 Illinois     -61.7%
 4 Florida     -59.0%
 5 Kentucky    -57.1%
 6 New Jersey     -55.9%
 7 Michigan    -54.5%
 8 Oregon      -53.9%
 9 Maryland    -53.2%
10 Texas                  -50.1%

These are the ten states with increased violent crime rates or else the least improvement.

Worse Rates or Least Improvement
 1 North Dakota       230.3%
 2 Montana           73.8%
 3 South Dakota         63.8%
 4 New Hampshire  50.9%
 5 West Virginia         36.5%
 6 Vermont          5.6%
 7 Wisconsin           2.0%
 8 Hawaii           -3.7%
 9 Maine              -6.4%
10 Alaska       -11.0%

Because the previous map was so busy, I decided to show one which only highlights the states with the top ten best and worst performances.

Border States.

The first thing that jumped out at me were the border states. There are four states which border Mexico: California, Arizona, New Mexico and Texas. Two of these, California and Texas, are among the top ten for the greatest improvement in violent crime statistics. Arizona and New Mexico have also seen healthy improvements of -40.8% and -35.6%, respectively.

There are ten states sharing land borders with Canada (eleven including Michigan, connected by bridges). Of these, six are among the bottom ten regarding improvement in violent crime rates.

So, what's going on here?

Hypothesis one
. There is something about being contiguous with Canada that raises crime rates. For example, perhaps border smuggling (narcotics, other criminal activities) is a real problem. I looked for support of this hypothesis but I came across only general reports and sensational articles rather than a comprehensive analysis.

In favor of this hypothesis.

As mentioned above, six out of ten is a surprisingly large number to achieve by chance.

Arguing against this hypothesis.

  • New York, a border state, had the highest drop in violent crimes. Minnesota, Michigan and Washington also had sizeable drops while Idaho had a modest drop. Michigan might be a separate case. It could be that with bridge crossings, Michigan has a well-regulated border. 
  • New Hampshire and Idaho have smallish borders.
  • This hypothesis doesn't explain West Virginia, South Dakota, Wisconsin or Hawaii. (But then does one hypothesis need to explain everything?)
  • The problems attendant with North Dakota's oil boom have been well-documented and are rather unique. The rise in violent crimes, however, preceded the population boom of the mid-2000s.

Alternative hypothesis. Most of these states have something else in common besides bordering Canada.

My original observation was that the District of Columbia had a precipitous fall in murders. Compared to the 50 states, Washington, DC is the most urban (100%). In contrast, Maine, New Hampshire, Vermont, Montana and West Virginia are all among the most rural of states. My first alternative hypothesis is that violent crime rates are up in primarily rural states and down in urban states.

So, what is rural and what is urban? The United States Census Bureau pondered this question and came up with two definitions of urban and one for rural that they have used since 1990. From their website:

  • Urbanized Areas of 50,000 or more people;
  • Urban Clusters of at least 2,500 and less than 50,000 people.
  • "Rural" encompasses all population, housing, and territory not included within an urban area or cluster.

While living in cities of below 2,500 people (or not in cities at all) seemed to define rural, I decided that urban should counted as people in the urbanized areas of 50,000 or more.

So I tested this hypothesis by comparing the most poorly performing states using two standards:

Which states have the highest percentage in rural populations.
Which states have the lowest percentage in urbanized areas.

These numbers can be derived from individual state profiles from the Census Bureau (Table 2 in the linked example). Often rural and non-urban correspond to one another. In other cases, such as Alaska, Alaska is very much a non-urban state, but is the 33rd most rural.

Here is the ranking of the fifteen worse performers in the violent crime statistics, according to the ranking as being rural or least urban.

State     V. Crimes     Most Rural     Least Urban
North Dakota  1 10 13
Montana    2 5 4
South Dakota 3 7 6
New Hampshire  4 11 17
West Virginia 5 3 7
Vermont 6 2 1
Wisconsin 7 18 23
Hawaii  8 46 35
Maine 9 1 3
Alaska 10 33 10
Iowa 11 12 11
Tennessee 12 17 20
Nevada 13 48 42
Arkansas 14 6 8
Pennsylvania 15 31 34

Six out of the top ten of the worst performers were also among the ten most rural states, along with six out of ten of the least urban. West Virginia and South Dakota now fit into the explanation. Among the top ten, not fitting into either explanation are Hawaii and Wisconsin.

There has been a lot of news about rises in rural crime. I am reluctant to link to these reports, inasmuch as it seems to me to be a sort of confirmation bias.

The Decline of White Privilege.

A lot has also been written about the decline in white privilege, that is, as America has become more diverse, the relative power of being white in America has declined. I am not going to comment on the validity of the arguments, but it is hard to escape that several of the poorer performing states are among those that are least racially diverse. Among some people, the word "urban" connotes a mix of minorities. Here the most poorly performing states are compared to their ranking as being least diverse.

State     White, non-Hispanic
North Dakota  7
Montana    6
South Dakota 10
New Hampshire  4
West Virginia 3
Vermont 2
Wisconsin 12
Hawaii  51
Maine 1
Alaska 32
Iowa 5
Tennessee 37
Nevada 46
Arkansas 25
Pennsylvania 19

Now we have seven of the least racially diverse states making up the top ten, along with Iowa (number 11 poorest performer, 5th least diverse) also falling into the group, and Wisconsin nearby (7th poorest performer, 12th least racially diverse). Hawaii, the most racially diverse state, is anomalous.

So, What's the Verdict?

Frankly, rural, non-urban, near Canada and mostly white overlap with one another. It is hard to tease one from the other. Still I would submit that something worth looking into is going on here. Violent crimes, whether rural or urban, have victims. It is possible we are ignoring the dramatic increases in crime rates in Montana, South Dakota, New Hampshire, North Dakota, and West Virginia because they are lost when we look at overall crime rates.

Changes in individual states are probably strongly driven by economic forces, economic success in North Dakota and the collapse of the coal economy in West Virginia. Perhaps a change in rural economy in general is at the heart of the problem. It should be noted that notorious "rust-bucket" states and many of those that have lost considerable manufacturing jobs, did well in terms of decreasing crime.

It is possible that there is a perception of crime increasing everywhere among those who are witnessing local increases. Therefore rural states may be more susceptible to concerns about crime even though overall crime statistics have drastically decreased.

A Great Divide.

The United States has become more polarized. Some in the media play up the fear of city and this is embraced, often with an undertone of racism and superiority: big cities have crime, cronyism, and are corrupt with Boss Tweed-type politicians. In defense of their arguments they will cite the higher total crime rates although it is likely that the urban warfare seen in fictional crime shows is mixed into the view. Psychotic taxi drivers stalk the streets. Living in New York City would be horrible.

Another segment of the media depict rural America as backwards, often through the lens of a different brand of superiority. They see rural America as uncultured, intolerant and teeming with corrupt Huey Long-type politicians. They will cite how these states come in near last in markers of health, education, and income. Klansmen stalk the streets. Living in Alabama would be horrible.

Neither of these are true pictures. I've lived in the inner city and in small towns. Life goes on. Most folks are decent. There is still a large chance you will not meet with a violent crime in either place.

Interestingly, you are more likely to die from violence in rural areas. Violent accidents, particularly vehicle accidents, dominate death-by-violence statistics.

Additional Notes.

Violent crime rates versus changes in violent crime rates. To be clear, urban centers have higher violent crime rates than rural areas. Concentrating people together has this effect. However, both crime rates and changes in crime rates have stories to tell, and while the total crime in urban centers is well-discussed, I have focused on changes in violent crime rates because they tell an important additional story.

In [1992] North Dakota was 14 times safer than New York. In [2012] it was 34.9% safer.

The FBI provides stern warnings about the misuse of the Uniform Crime Reporting statistics. These warnings ask you not to take crime out of context (urbanization, poverty, etc.). These types of warnings are particularly directed to their statistics on crime in universities. In this analysis, I have tried to find context.

Don't Chase Blips.

Other than in the cases where violent crime rates steadily improve or worsen year-to-year, every year, you will have upturns and downturns. These get magnified in the press, sometimes even microblips such as single month reports in single cities are drawn out as meaningful. This is usually done with fear-mongering. Sometimes crime in one locality will go up dramatically, while in another it goes down. It either balances out and should be taken in this context, or else it will continue and become a trend. In the latter case, waiting for a multi-year change is necessary.

In line with this, I've endeavored to avoid the overtly political commentary that can come with this subject. I raised my eyebrows as to why two mostly liberal border states, Minnesota and New York have had drops in crime rates while more conservative border states did not. I didn't want to go there. What is conservative? What is liberal? Texas has certainly had more than two decades of conservative governors and has had a drastic drop in violent crime. California is a mixed bag of governance and had an even greater drop. There may be an overtly political story somewhere in these data, I didn't seek it out. It will probably not be a simplistic one.

A note for those who believe in conspiracies and doctored crime rate statistics. Yes, doctoring can occur. Some states change definitions. The FBI changes definitions over time. But to describe overall changes in this form is paranoia. There is no grand conspiracy that explains the decreases in crime rates in California, New York, Texas, Florida, etc. over a period of decades.

Continued in New York City Versus New York State.

Martin Hill Ortiz is the author of Never Kill A Friend, Ransom Note Press.

Never Kill A Friend, Ransom Note Press

Never Kill A Friend is available for purchase in hard cover format and as an ebook.
The story follows Shelley Krieg, an African-American detective for the Washington DC Metro PD as she tries to undo a wrong which sent an innocent teenager to prison.

Hard cover: Amazon US
Kindle: Amazon US
Hard cover: Amazon UK
Kindle: Amazon UK
Barnes and Noble 


  1. Lots of adjectives come to mind such as stunning, fascinating, interesting, valuable, worthwhile. I'm really pleased to see what appears to be a thoughtful, careful, objective accurate presentation. It is also surprising and useful. Thank you so much!

  2. Thanks for your comments. The take-away message here is: something important is going on.
    Rural heroin use is up. This sort of headline can be found: The New Heroin Epidemic.
    The Atlantic talks about heroin replacing legal opioids in rural America.

  3. I am so impressed with your research. Clearly something is going on. But why really impressed me is that you were very clear about what you were measuring and that played against the larger picture. This is rational, exacting reporting and shows what data can do when it's not being distorted or manipulated to prove a point. Awesome job!

  4. I am so impressed with your research. Clearly something is going on. But why really impressed me is that you were very clear about what you were measuring and that played against the larger picture. This is rational, exacting reporting and shows what data can do when it's not being distorted or manipulated to prove a point. Awesome job!